export performance of central and eastern european countries macro and micro fundamentals

The performance of high percentiles firms drives external results: more sophisticated and higher volumes of exports are recorded in regions where the productivity distributions have fatt

Procedia - Social and Behavioral Sciences 195 ( 2015 ) 514 – 523

Available online at www.sciencedirect.com

1877-0428 © 2015 The Authors Published by Elsevier Ltd This is an open access article under the CC BY-NC-ND license

( http://creativecommons.org/licenses/by-nc-nd/4.0/ ).

Peer-review under responsibility of Istanbul Univeristy.

doi: 10.1016/j.sbspro.2015.06.260

ScienceDirect

World Conference on Technology, Innovation and Entrepreneurship

Export performance of Central and Eastern European Countries:

Macro and Micro Fundamentals

a Bucharest University of Economic Studies, Bucharest, Romania

Abstract

This paper analyses the evolution of exports’ value and sophistication during 2004-2012, using Hausmann, Hwang & Rodrik (2007) methodology Focusing on Central and Eastern Europe, it is underlined that these states have to intensify efforts to support innovation-led growth through higher investments into technology intensive and sophisticated sectors Export performance indicators are heterogeneous across countries, but also within the same country We use panel fixed-effects models in order to investigate regional disparities of external results at county level in the case of Romania The empirical results show that counties’ trade volumes and sophistication are explained by region specific factors (value added and foreign direct investment) and, more importantly, by micro-level behaviour Based on Total Factor Productivity, computed by Wooldridge (2009) GMM method, we find that heterogeneity of firm-level technology and efficiency is the key for explaining the differences in aggregate trade outcomes The performance of high percentiles firms drives external results: more sophisticated and higher volumes of exports are recorded in regions where the productivity distributions have fatter right tails This evidence brings into question the efficiency of policy measures targeting external competitiveness, highlighting the importance of taking into account the entire firm-level performance distributions, rather than just the average

© 2015 The Authors Published by Elsevier Ltd

Peer-review under responsibility of Istanbul University

Keywords: sophistication; TFP; heterogeneity; exports; firm-level data

* Corresponding author Tel.: +40-724-414-025

E-mail address: ana02mary@yahoo.com

© 2015 The Authors Published by Elsevier Ltd This is an open access article under the CC BY-NC-ND license

( http://creativecommons.org/licenses/by-nc-nd/4.0/ ).

Peer-review under responsibility of Istanbul Univeristy.

1 Introduction and state of art

Investigating the determinant factors for countries’ divergent performances on external markets has become one

of the main interests for policy makers and academics, as an in-depth understanding of the engines that enable a country’s firms to successfully compete in international environment is essential in adequately formulate policy responses to address external competitiveness issues However, active participation on export markets does not guarantee a positive impact on domestic living standards Recent empirical evidence (Hausmann, Hwang &Rodrik

2007) highlights that not exporting per se contributes to improved economic conditions, but what matters for growth

is exports’ structure, in terms of productivity level associated with the specialization pattern (implicitly, embedded technology, quality and innovation of exported goods)

The macroeconomic fundamentals of exports’ dynamics and degree of sophistication or diversification have been highly debated In Heckscher-Ohlin model (1933), factor endowments and comparative advantages are the main triggers of countries’ exports specialization pattern Krugman (1983) introduced the so called new trade theory, which treats imperfect competition, economies of scales, product differentiation and intra-industry trade Recent literature (Harding & Javorcik 2009; Xu & Lu 2009; Weldemicael 2012 etc.) highlighted that human capital, macroeconomic environment, institutional framework and foreign direct investment are among the most important determinants of exports’ sophistication level However, macro approach of competiveness, based on a representative agent, was proved to be less satisfactory, given the high firm-level heterogeneity (Bernard et al 2011) In order to properly assess a region’s competitive position, complementing traditional aggregate macro indicators with microeconomic foundations is needed As such, the centre of the international trade studies started shifting from countries and industries towards firms and products Models introduced by Bernard et al (2003) or Melitz (2003) focus on firm heterogeneity, resource reallocation and differences between trading and non-trading firms All these aspects have significant implications for external trade and aggregate productivity growth The Competitiveness Research Network (2014) shows that micro-level data contributes to a better understanding of external competitiveness, as firm heterogeneity hides different behaviours related to productivity and unit labour costs

We make two contributions to the literature treating firm heterogeneity and exports’ determinants Firstly, we analyse the recent developments in exports’ volumes and sophistication level for a sample of important Central and Eastern Europe (CEE) countries: Bulgaria, Czech Republic, Estonia, Hungary, Latvia, Lithuania, Poland, Romania, Slovak Republic and Slovenia In line with most of the abovementioned studies on export sophistication (which were however mainly focused on China), we use Hausmann, Hwang & Rodrik (2007) methodology for assessing sophistication indicators

Secondly, our study examines the variation of trade outcomes across regions within the same country – Romania,

an important economy in CEE in terms of GDP, market size, exports’ volumes etc We examine the link between export results at county level and the shape of firms’ productivity distributions The novelty of this study stems from explaining not only exports’ volumes, but also their degree of sophistication, by micro-level Total Factor Productivity (TFP), accounting for differences in technology and efficiency among companies In constructing TFP,

we used the GMM estimator proposed by Wooldridge (2009) It is shown that micro-level performance indicators’ distribution influences county external competitiveness, with impact on country’s overall export dynamics (in the aftermath of the crisis, the exports recovery was driven by a small number of counties, mainly in the superior classes regarding export sophistication)

The paper is organized as follows In Section 2 we present the methodology for computing exports’ sophistication indicators and firm-level productivity, while in Section 3, we describe the data used in the empirical exercise In Section 4 we present the results related to CEE competitiveness evolution, regional disparities of export performances at county level in case of Romania and the impact of firm-level variables on aggregate trade outcomes Concluding remarks and possible policy options are formulated in Section 5

516 Ana-Maria Cazacu (Bancu) / Procedia - Social and Behavioral Sciences 195 ( 2015 ) 514 – 523

2 Methodology

The methodology allows the computation of sophistication indicators for countries worldwide, with particular focus on CEE Moreover, we investigate regional differences in export performances within one of the most important CEE country, namely Romania Exports’ results at county level are explained by aggregate indicators and characteristics of firms activating in the respective area (more specific, the shape of micro-level productivity distributions) Estimations were done by panel fixed effects models at county level

2.1.Export sophistication indicators

Hausmann, Hwang & Rodrik (2007) developed an index capturing the income level of countries’ exports and underlined its role for economic growth Computation of export sophistication (EXPY) of a country starts from productivity equilibrium in a modern sector The novelty of the abovementioned study derives from ranking commodities by their productivity level (PRODY)

PRODY (Equation (1)) is a weighted average of the per-capita GDPs of countries exporting a certain product, where the weights reflect the revealed comparative advantage of each country The higher the PRODY index (and thus the income level) associated with a product, the more sophisticated it is considered to be This is linked to factor endowment theory, stating that developed countries export more capital-intensive goods, characterized by higher productivity levels, while poor countries mainly specialize in natural resources or labour-intensive products The productivity level that corresponds to a country’s export pattern (EXPY) is then computed as a weighted average of the PRODY corresponding to each product traded abroad by the respective country The weights are represented by the share of the product’s exports in the country’s total exports

¦ ¦j

j

j jk j

j jk

X x

X x PRODY

) / (

) / (

¦

k

k j

jk

X

x



where Y is the GDP per-capita of country j, k denotes the traded good, x the export value of country j for product k and X is the total exports’ volume of country j

2.2 Firm-level estimation of productivity indicators

In order to investigate the impact of micro-level characteristics on aggregate trade outcomes, we have chosen as firms’ performance indicator a measure accounting for technology, employees’ knowledge (human capital) etc.: TFP Deflating value added for each company in a certain industry was done based on deflators at NACE rev 2, 2 digits level, provided by Eurostat database TFP, one of the most important firm-level competitiveness indicators, indicates how efficiently and intensely the inputs are utilized in production (European Commission [2011, 399]) and can be defined as the portion of output not explained by the inputs used in production For estimating TFP, we started from the production function for each firm in the economy, which is assumed to be of Cobb-Douglas type:

it it it m it l it k

where yit is the real value added from firm i at time t, t 2007:2012 , k itcapital input, l it labour, m it materials,

0

E the sector-specific intercept, Ek, Eland Em the output elasticity at changes in capital, labour and intermediate inputs respectively, v it unobserved productivity, and e it a sequence of shocks

As TFP is an unobservable variable, for its estimation we used the approaches proposed by Levinshon & Petrin (2003) and Wooldridge (2009) These approaches treat the endogeneity that affects OLS estimators, resulted from the correlation between firm-level productivity and the choice of inputs Levinshon & Petrin (2003) use intermediate inputs as proxy for productivity shocks, in a two-step semi-parametric approach (modifying the estimator proposed

by Olley & Pakes 1996) Wooldridge (2009) proposed estimating TFP in a GMM framework, based on the assumptions that i) the error terms are conditional mean independent of current and past inputs (Equation (4)), while ii) productivity innovations are correlated with variable inputs (l it,m it), but uncorrelated with the state variable (k it) and all past values of the inputs (Equation (5))

T t m l k m l k

m

l

k

e

E[ it| it,it, it, it1,it1, it1 i1,i1, i1] 0, 1:

( 4 )

T t v v E m l k m l

k

k

v

E[ it| it, it1,it1, it1 i1,i1, i1] [ it| it1], 2: 



( 5 )

The production functions for each firm are estimated assuming that labour and capital elasticity are the same within a certain industry (defined as NACE rev 2, 2 digits) Lagged values of variable inputs are used as instruments for capital TFP is obtained for each firm as a residual in the production function, indicating differences

in efficiency of using inputs in production process Both approaches used for estimation (Levinshon & Petrin 2003 and Wooldridge 2009) deliver similar production function coefficients and thus TFP In consequence, in the rest of the paper, I will only present the results based on Wooldridge (2009) After computing TFP for each firm, the empirical distributions of productivities are generated for each county Of particular interest in the empirical process are distributional moments like mean, skewness, median and higher order percentiles (75th and 90th percentiles), which will be used to explain aggregate trade outcomes For robustness check, the regressions were also run using labour productivity (real value added generated by each company divided by number of employees)

3 Data used for assessing aggregate export performances and firm-level competitiveness

The databases at hand comprise two types of information: (1) macroeconomic indicators for a wide variety of countries in the world and (2) in case of Romania, a rich dataset on micro-level financial and trade data

x Information on trade flows and GDP per capita of exporting countries is used for assessing the productivity levels associated with traded products and the sophistication of countries’ exports (following Hausmann, Hwang & Rodrik 2007 methodology, described in Section 2.1) The trade data in current USD is provided by United Nations Commodity Trade Statistics Database, covering 260 products at 3-digit level, according to SITC rev 3 World Bank World Development Indicators is the database for per-capita GDP, expressed in constant 2005 USD The dataset refers to a total of 175 countries worldwide, covering 2004-2012 period, with annual frequency Considering a longer time horizon could make difficult avoiding structural breaks in trade data, due to capital account liberalization episodes For example, in CEE countries, the bulk of liberalization took place between 1995-2003 (Arvai 2005) In constructing PRODY for each of the 260 products analysed, we used PPP-adjusted GDP for each exporting country in 2004-2012 In order to further compute EXPY (Equation (2)) average PRODY for the entire period was used By holding PRODY for all products constant, we aim to capture countries’ export structure shifts toward more or less sophisticated products, rather than changes in development level, as indicated by evolution of GDP per-capita For Romania, sophistication indicators are computed at a higher disaggregation level, some authors considering this approach might lead to a better quality of the results (Péridy & Jouini 2013) PRODY is computed for each product classified according to 6-digits HS71, while EXPY is analysed for each county, in order to detect

1 Data for each country in the world on exports per product classified according to HS7, 6 digits level, is extracted from United Nations Commodity Trade Statistics Database for 2012 Due to computational considerations, we choose to use PRODY and EXPY computed based on 3 digits SITC rev 3 classification for 2004-2012 As checked in the case of Romania, there are no significant differences between EXPY levels

518 Ana-Maria Cazacu (Bancu) / Procedia - Social and Behavioral Sciences 195 ( 2015 ) 514 – 523

whether aggregate country performance hides significant territorial disparities of export performances At economy-wide level, the difference between EXPY based on SITC rev 3 and HS7 is small

x For Romania, data on external trade is provided by National Institute of Statistics (NIS), with detailed information on exported products, classified according to Harmonized System 7 (HS7), 6 digits level National Trade Register Office is the source for foreign direct investments Firm balance sheet information comes from Ministry of Public Finance, covering 2007-2012 period The balance sheet data is used for estimating unobservable TFP at industry level (as detailed in Section 2.2) All variables (Table 1) are in logarithmic form and expressed in real terms Value added is deflated with deflators from Eurostat National accounts at NACE rev 2 level, while capital and turnover are deflated by the GDP deflator Materials are deflated with the deflator corresponding to sector 35 in NACE REV.2 (electricity, gas, steam and air conditioning supply)

Table 1 Definitions of firm-level variables used in estimating production functions Variable Description

value added (RON) turnover adjusted for changes in inventories and capitalized production minus intermediate costs

capital (RON) tangible fixed assets

labour (persons) number of employees, full-time equivalent, yearly average

materials (RON) raw materials, supplies, other materials expenses, utilities, cost of merchandise

The data was cleaned by excluding negative value added companies, with small impact on sample representativeness All the estimations were done after trimming outliers, defined as values lower than 1st percentile

of the variable or higher than 99th percentile The outlier treatment does not significantly affect the main results, these being robust to different definitions for outliers (values lower than 5th percentile or above 95th percentile of a variable) or procedures for treating them (for example winsorising) The final unbalanced panel of almost 730,000 companies accounts for over 98% of the labour force and turnover in the economy (as of 2012) and covers 83% of active firms during 2007-2012 period (Table 2)

Table 2 Final sample used in firm-level productivity estimation

Bucharest-Ilfov 174,658

North-East (Moldova) 79,392

South (Muntenia) 79,337

South-East (Dobrogea) 85,693

South-West (Oltenia) 53,430

4 Results and discussions

4.1 Evolution of export performance

The exports’ dynamics accelerated during 2004-2012 for almost all countries in the sample, due to trade reforms and liberalization processes, increased number of tradable goods and higher integration in global production chains Although an increasing trend was also recorded for sophistication levels, differences between countries still persist The rankings confirm previous findings (Hausmann, Hwang & Rodrik, 2007; Anand, Saurabh & Spatafora, 2012),

based on the two classifications

which correlate exports’ structure and economic development Among countries with the highest EXPY value, we

meet Ireland, Switzerland, Japan etc (Table 3), due to high involvement of exporting sophisticated products (such as

pharmaceuticals, clocks, vehicles and engines, medical equipment etc.) On the other hand, least developed economies, especially from Africa, are at the bottom of world ranking, exporting mainly primary products (textiles,

seeds, spices, cocoa, coffee)

Table 3 Countries’ ranking according to EXPY (2012) Top 10 countries according to EXPY EXPY (USD) Bottom 10 countries according to EXPY EXPY (USD)

Ireland 23,203 Sao Tome and Principe 3,812

At European level, most CEE countries are lagging behind top performers as Ireland, Finland, Luxembourg and

Germany However, the CEE countries of interest increased their export market shares in 2004-2012 (with 0.4 percentage points, to 3.7% in world exports) Also, they shifted towards trading more sophisticated products, recording a higher growth rate of EXPY (5% average increase in 2004-2012, compared to 1% in EU) Since 2004,

the gap between CEE and European average decreased and nine of 10 of the abovementioned countries recorded an

upgrade in world rankings according to EXPY index (Table 4) The situation is similar for Romania, in which case

EXPY increased by 20% in 2012 compared to 2004 (this leading to a gain of 29 places in world ranking in the same

period) However, Romania still has an unfavourable position among its peers, being placed in the lower half of the

region outperforming only Bulgaria, Latvia and Lithuania

Table 4 Ranking of selected CEE countries according to EXPY index

Czech Republic 19 17

Slovak Republic 31 25

520 Ana-Maria Cazacu (Bancu) / Procedia - Social and Behavioral Sciences 195 ( 2015 ) 514 – 523

4.2 Export performance at county level in Romania

After obtaining a flavour of the regional developments, we investigate the territorial asymmetry related to exports’ volumes and sophistication within a single country, namely Romania The empirical results show that i) exporting activity is concentrated at a small number of counties and ii) there is a substantial EXPY variation among regions Although during 2007-2012 most of the 42 Romanian counties increased their export volumes, exports’ volume is characterised by a high degree of concentration: as of 2012, the first 10% of counties account for 39% of exports’ volumes Regarding exports’ sophistication, there have been recorded divergent evolutions: 45% of counties registered a decrease in EXPY, while the rest knew a favourable evolution of the index

At the top of the counties exporting goods with a high degree of sophistication there are generally richer regions

in terms of GDP per-capita, i.e Central and Western counties (Braúov, Sibiu etc.), Figure 1 The capital, Bucureúti, has an export sophistication index close to the average, despite being the most developed region At the same time, some counties as Iaúi and Olt have a surprisingly high level of sophistication, despite a below average GDP per-capita This situation is driven by the large share of export sales of nuclear reactors, boilers, machinery and mechanical appliances (oil pumps or other nuclear parts) and new pneumatic rubber tyres for motor cars, respectively

Fig 1 EXPY variation of Romanian counties

A vulnerability of Romanian activity is given by the persistence of a relatively low share of counties which export products with a high degree of sophistication: counties situated above 75th percentile in terms of EXPY accounted for about one third of total export volume in 2012 Increasing the export activity of these counties would positively impact Romania’s aggregate export performance, which could further translate in increased competitiveness capacity and growth Using the comprehensive firm-level database, we proceed at identifying the fundamentals that enable different regions of Romania to have better export performance than others The link between export outcomes for each county on one hand and county characteristics and firm-level competitiveness indicators on the other hand is investigated, in a panel fixed-effects model framework at county level:

it it it it it

where T is a measure of trade performance (EXPY or exports’ value in national currency equivalent) of county i in year t, GVA denotes gross value added generated in each county (as a sum of value added generated by each firm in the county), FDI denotes the total value of foreign direct investment directed to firms activating in the respective region, c it accounts for county fixed-effects (confirmed by Hausman 1978 test) and Y is vector of moments of firms’ TFP distribution (mean, median, skewness, 75th percentile, 90th percentile), as described in Section 2.2 All regressions were estimated correcting for errors’ heteroskedasticity, as indicated by modified Wald test for groupwise heteroskedasticity in the residuals of a fixed effect regression model As shown in Table 5, both exports’ volumes (col (1)-(4)) and level of sophistication (col (5)-(8)) are higher in counties that generate higher value added and benefit from larger foreign direct investment Thus, increasing Romania’s attractiveness for foreign investors is essential for improving international competitiveness Even after controlling for the two abovementioned indicators of counties’ development and other unobserved, time invariant, fixed effects, it is shown that superior export performances are recorded in regions where firms with higher average productivity activate However, the average productivity might not representative for the underlying distribution of firms operating in a certain territory, especially taking into account the high degree of heterogeneity among companies highlighted by the productivity distribution at economy-level (Figure 2) Productivity distributions (for both labour productivity and TFP) are far from normal, displaying strong right asymmetry: there are only a few companies with high performances, as opposed to low productivity companies, which are much more frequent The significant difference between average productivity and the median of the distribution underlines the importance of taking into account the empirical findings on the underlying distribution of firms’ performance indicators, which is not reflected by simply averaging The heterogeneity among economic agents is significant, regardless they activate in the same industry or region

0 20,000

40,000

60,000

80,000

100,000

120,000

140,000

RON/employee

0 5,000 10,000 15,000 20,000 25,000 30,000

Figure 2 Distributions of firm-level labour productivity (left) and TFP (right) in the Romanian economy

Note: The bottom and up lines represent 10 th and 99 th percentile respective The bars indicate interquartile range (between 25 th and 75 th percentile)

The role of the entire shape of the micro-level distribution in explaining aggregate trade outcomes is also confirmed by the positive and significant coefficients of levels of productivity at the top of the distribution (75th and

90th percentiles) The skewness is also positively linked with external results As positive skewness indicates how much the distribution is biased towards higher productivity levels (how fat is the right tail), the regressions’ results highlight that higher frequency in the distribution of high productivity firms improves trade performance As such, two regions with similar average productivity might have different performances depending on the right tail of the distribution, as -most likely- the performance of high percentiles drives aggregate trade outcomes, regardless the mean The results are consistent for both productivity measures (labour productivity or TFP), or when median instead of mean productivity is used As top percentiles firms are shown to be the engines that enable better external

522 Ana-Maria Cazacu (Bancu) / Procedia - Social and Behavioral Sciences 195 ( 2015 ) 514 – 523

trade results, strategies promoting export performance have to be based on a deeper analysis of micro behaviour

The regressions’ result highlight that the aggregate results and, implicitly, the impact of macroeconomic policies, are

significantly affected by the underlying distribution of firms Counties’ performances are driven by a small, highly

productive set of companies and understanding their behaviour should be the centre of policy makers More specific,

knowing the characteristics (in terms of size, ownership, age etc.) and dynamics of firms situated in different parts

of productivity distribution in a certain region, could prove essential in evaluating the impact of particular policies

and shocks on different group of firms

Table 5 Export performance at county level and firm productivity

GVA 0.523*** 0.509*** 0.659*** 0.382** 0.307*** 0.303*** 0.334*** 0.138**

(0.115) (0.121) (0.127) (0.161) (0.0912) (0.0905) (0.0924) (0.0674) FDI 0.0825** 0.0794* 0.106*** 0.0678 0.0653** 0.0644** 0.0740*** 0.0462**

(0.0385) (0.0407) (0.0360) (0.0466) (0.0251) (0.0247) (0.0249) (0.0192) TFP_avg 0.598*** 0.663*** 0.371*** 0.389***

(0.146) (0.178) (0.0932) (0.0971)

County Fixed Effects Yes Yes Yes Yes Yes Yes Yes Yes

R-squared 0.359 0.368 0.318 0.379 0.460 0.458 0.448 0.551

Col (1)-(4) Dependent variable: exports’ value Col (5)-(8) Dependent variable: export sophistication index All variables are in logs

Robust standard errors in parentheses *** indicates significance at 1%; ** significance at 5%; * significance at 10%

5 Conclusions Policy options

Based on the analysis of aggregate exports’ dynamics in terms of volumes and sophistication, this paper

highlighted the global increasing trend in these indicators in the last years, along with a high heterogeneity across

countries’ performances Using aggregate sophistication indicators developed by Hausmann, Hwang & Rodrik

(2007), it is underlined that CEE countries, including Romania, have to intensify efforts to support innovation-led

growth through higher investments into knowledge and technology-intensive sectors Increasing the level of exports’

sophistication would enhance competitiveness and further economic growth, as recent studies emphasized that

sophisticated sectors are more likely to act as an accelerator for economic development (Anand, Saurabh &

Spatafora 2012; Jarreau & Poncet 2012) Identifying the fundamentals for disparities in trade outcomes is of

particular importance for policy options and strategies aiming at improving external performances and ultimately

economic growth The present study highlights the significant impact of firm-level characteristics (mainly related to

differences in technology, human capital etc.) on aggregate trade outcomes It is shown that valuable explanations

for external trade results -apart from regions’ specificities- can be derived from the shape of performance indicators

(namely TFP) of companies activating in the respective area Exports value is not only concentrated at a small

number of firms, but even more, its evolution is determined by top performers Targeting the average firm in a

region in order to improve external trade results could lead to unsatisfactory results, since, in reality, firms close to

average have a very low frequency A solution would be adopting tailor made measures, adapted at the needs of

firms in different clusters of efficiency, in order to shift the entire performance distribution to the right, rather than addressing increases in average indicators At the same time, policy makers could target regions where the right tail

of firm-level productivity distributions could become thicker, as firms which have already superior performances compared to the average firm in the economy could transform in engines which propagate positive effects for other companies in the production chain As there is still room for better understanding of micro transmission mechanism, the evidence provided by the present paper underlines the need to properly assess the competitive position of economies or regions by complementing the analysis of macroeconomic developments with insights on firm-level behaviour

References

Altomonte, C., Navaretti, G., di Mauro, F & Ottaviano, G (2011) Assessing competitiveness How firm-level data can help Bruegel Policy

Contribution Issue 16

Anand, R., Saurabh, M & Spatafora, N (2012) Structural Transformation and the Sophistication of Production IMF Working Paper No 12/59 Arvai, Z (2005) Capital Account Liberalization, Capital Flow Patterns and Policy Responses in the EU’s New Member States IMF Working

Paper No 213

Bernard, A.B., Eaton, J., Jensen, J.B & Kortum, S (2003) Plants and Productivity in International Trade American Economic Review

93(4):1268.1290

Bernard A., Jensen, J.B Redding, S & Schott, P (2011) The Empirics of Firm Heterogeneity & International Trade Annual Review of

Economics, Vol 4(1), 283-313, 07

Dunaway, S (2009) Global Imbalances and the Financial Crisis The Council on Foreign Relations, Special Report No 44

European Commission (2011) Innovation Union Competitiveness report 2011, European Commission

Eurostat (2012) Data base [online] Available at: <http://epp.eurostat.ec.europa.eu/portal/page/portal/statistics/themes> [Accessed in May 2014]

Flam, H & Flanders, M.J (1991) Heckscher-Ohlin Trade Theory Cambridge, MIT Press

Harding, T & Javorcik, B.S (2009) A Touch of Sophistication: FDI & Unit Values of Exports CESIFO Working Paper No 2865

Hausman, J A (1978) Specification Tests in Econometrics Econometrica 46 (6), 1251–1271

Hausmann, R., Hwang, J & Rodrik, D (2007) What You Export Matters Journal of Economic Growth 12(1): 1–25

Jarreau, J & Poncet, S (2012) Export sophistication & economic growth: Evidence from China Journal of Development Economics, Vol 97,

Issue 2, 281-292

Krugman, P (1983) New theories of trade among industrial countries The American Economic Review, 73/2, 343-347

Levinsohn, J & Petrin, A (2003) Estimating Production Function Using Inputs to Control for Unobservable Review of Economic Studies 70,

317–341

Melitz, M.J (2003) The Impact of Trade on Intra-Industry Reallocations & Aggregate Industry Productivity Econometrica, Vol 71(6),

1695-1725

Péridy, N & Jouini, N (2013) Diversification and sophistication as a lever for the structural transformation of North African economies United

Nations, Economic Commission for Africa, draft report

Olley, G.S & Pakes, A (1996) The Dynamic of Productivity in the Telecommunication Industry Econometrica 64(6), 1263–97

Rodrik, D (2006) What's So Special about China's Exports?, China and World Economy Institute of World Economics & Politics, Chinese

Academy of Social Sciences, vol 14(5), 1-19

The Competitiveness Research Network (2014) Micro-based evidence of EU competitiveness The CompNet database ECB Working Paper No

1634

The Stata Journal (2001) Residual diagnostics for cross-section time series regression models Stata Journal No 1, 101–104

Weldemicael, E.O (2012) Determinants of Export Sophistication Melbourne University, mimeo

Wooldridge, J.M (2009) On estimating firm-level production functions using proxy variables to control for unobservables Economic Letters,

Vol 104, Issue 3, 112-114

World Bank (2012) World Development Indicators, Data base [online] Available at: <http://data.worldbank.org/products/wdi> [Accessed in March 2014]

World Trade Organization (2013) World Trade Report, [online] Available at: <http://www.wto.org/english/res_e/publications_e/wtr13_e.htm> [Accessed in June 2014]

Xu, B & Lu, J (2009), Foreign direct investment, processing trade, & the sophistication of China's exports China Economic Review, 20(2009),

425í439

Xu, B (2010) The sophistication of exports: Is China special? China Economic Review, 21 (2010), 482–493

United Nations Commodity Trade Statistics Database (2012) Data base [online] Available at: <http://comtrade.un.org/data/> [Accessed in March 2014]